fix parsing db and tests

This commit is contained in:
qananasikq
2026-04-08 14:23:50 +03:00
parent fd1d1d3c9f
commit 3edd706110
7 changed files with 112 additions and 13 deletions

View File

@@ -1,3 +1,4 @@
import json
import logging
from contextlib import contextmanager
from datetime import datetime, timezone
@@ -13,6 +14,41 @@ from .schemas import CarRecord
logger = logging.getLogger("iaai_scraper.db")
CAR_DB_FIELDS = {
"parser_id",
"brand",
"model",
"year",
"price",
"currency",
"mileage",
"country",
"is_sold",
"color",
"drive",
"gearbox",
"steering_wheel",
"body_type",
"engine_volume",
"selling_type",
"one_owner",
"new_car",
"is_hidden",
"origin",
"origin_url",
"origin_id",
"is_damaged",
"evaluation",
"non_smoking",
"rental",
"repair_history",
"slug",
"last_seen_at",
"content_hash",
"raw_attributes",
}
class PersistenceService:
def __init__(self, settings: Settings) -> None:
@@ -64,15 +100,21 @@ class PersistenceService:
for image_payload in images:
session.add(Image(fullres_image=str(image_payload["fullres_image"]), preview_image=str(image_payload["preview_image"]), order_index=int(image_payload.get("order_index", 0)), car_id=car_id))
@staticmethod
def _car_payload(record: CarRecord) -> dict[str, object]:
payload = record.model_dump(mode="python")
result = {key: value for key, value in payload.items() if key in CAR_DB_FIELDS}
# Serialize raw_attributes dict to JSON string for Text column.
if "raw_attributes" in result and isinstance(result["raw_attributes"], dict):
result["raw_attributes"] = json.dumps(result["raw_attributes"], ensure_ascii=False, default=str)
return result
def upsert_car(self, record: CarRecord):
"""Insert/update/skip по content_hash."""
# Готовим payload отдельно от вложенных изображений и служебных полей.
payload = record.model_dump(mode="python")
images = payload.pop("images", [])
payload.pop("raw_attributes", None)
payload.pop("mapping_notes", None)
content_hash = payload.pop("content_hash", "")
payload["content_hash"] = content_hash
# В БД отправляем только поля, реально существующие в финальной схеме cars.
payload = self._car_payload(record)
images = [image.model_dump(mode="python") for image in record.images]
content_hash = str(payload.get("content_hash") or "")
with self.session_scope() as session:
# поиск по origin_id
car = session.execute(select(Car).where(Car.origin_id == record.origin_id)).scalar_one_or_none()